I am trying to code an algorithm that acts a bit like Amazon\'s \"People who bought this also bought\".
The difference between the two is that mine is just counting the
Ok. I think i figured it out. part of the work is code implementation.
What i did was to group by the sessionID, productID.
then, in my code I iterate thought each sessionID, and I make a dictionary with pairs.
for instance, if I have pid 10 and 20 and 30, which are a clique basically.
so I insert to the dictionary in the following way:
1. 10-20, weight 1
2. 20-10, weight 1
3. 10-30, weight 1
4. 30-10, weight 1
5. 20-30, eight 1.
6. 30-20, weight 1.
in case I encounter one of the values again, i will add +1 to the appropriate pair/s.
at the end, i will have the weights and pairs aligned.
all I have to do now is by a given productID, to scan the table and find the clique it is inside.
if you have improvements suggestions, please let me know!
thanks all!